full-train-mert-decode.test.svn-base

moses开源的机器翻译系统

#!/bin/bash

# $Id: consolidate-training-data.perl 40 2007-03-14 22:48:03Z hieu $

function die() {
  echo "$@"
  exit 1
}
RUNDIR=`pwd 2> /dev/null`
if [ -z $RUNDIR ]; then
  RUNDIR=`pwd`
fi;


[ -d $WORKSPACE ] || die "Missing $WORKSPACE"
echo "Workspace: $WORKSPACE"
echo "This test is expected to take an hour."

MOSES=$WORKSPACE/moses-cmd/src/moses

[ -x $MOSES ] || die "Missing $MOSES"

export SCRIPTS_ROOTDIR=$WORKSPACE/scripts

cp -r ../cs-en-sample ./corpus || die "Missing $RUNDIR/../cs-en-sample"
echo "Copied cs-en-sample files"

mv corpus/test.* ./ || die "Missing corpus/test.*"
cp corpus/train.cs ./dev.src || die "Missing corpus/train.cs"
cp corpus/train.en ./dev.ref || die "Missing corpus/train.en"

echo "Test and 'tuning' data ready. Tuning equals training."

bindir=$WORKSPACE/bin
[ -d $bindir ] || die "Please create $WORKSPACE/bin and put GIZA++ and such there"

echo "Starting training script."

$SCRIPTS_ROOTDIR/training/train-factored-phrase-model.perl \
  --bin-dir=$bindir \
  --f cs --e en \
  --translation-factors 0-0 \
  --decoding-steps t0 \
  --first-step 1 \
  --last-step 9 \
  --corpus corpus/train \
  --root   . \
  --parallel \
  --lm 0:3:$RUNDIR/corpus/lm.en.gz \
  || die "Failed to train the model"

echo "Finished moses.ini, merting."

$SCRIPTS_ROOTDIR/training/mert-moses.pl  \
    --working-dir=mert-tuning \
    $RUNDIR/dev.src \
    $RUNDIR/dev.ref \
    $MOSES \
    $RUNDIR/model/moses.ini \
    --decoder-flags="-dl 6 " \
    || die "Merting failed"

# Typical parallelization flags
#    --jobs=10 \
#    --old-sge \
#    --queue-flags=" " 

echo "Finished merting, filtering phrases for test data."

$SCRIPTS_ROOTDIR/training/filter-model-given-input.pl \
  filtered-opt \
  mert-tuning/moses.ini \
  test.src \
  || die "Failed to filter phrases for test data."

echo "Finished filtering. Deconding test set."

$MOSES \
  -dl 6 \
  -f filtered-opt/moses.ini \
  -i test.src \
  > test.opt \
  || die "Decoding of test set failed."

$SCRIPTS_ROOTDIR/generic/multi-bleu.perl test.ref \
  < test.opt > test.opt.bleu \
  || die "Final scoring failed"

# Typical parallel decoding:
#	$SCRIPTS_ROOTDIR/generic/moses-parallel.pl \
#	  -decoder $MOSES \
#	  -jobs 10 \
#	  -queue-parameters " " \
#	  -old-sge \
#	  -inputfile ./test.src \
#	  -config ./filtered-opt/moses.ini \
#	  > test.opt.out

echo "Success."